It seems that the majority of Big Data/Data Science applications are designed to give advertisers insight into things I don't really want them to have insight into. That really sucks, because the technology is cool, but I don't want to help build that kind of future. It's kind of analogous to how I feel about Computer Vision: there are a handful of legitimate purposes for it, but most applications of the technology fall somewhere between "I don't like that idea" to "that's totally unethical".
There are plenty of Big Data applications that aren't unethical. Back in the 90s (before the term Big Data was conceived) two of the biggest users of Teradata were P&G and Wal*mart. It was more about supply chain and retail store efficiency than anything nefarious. Big data helped make sure that store shelves had what people wanted.
Today there are mass spamvertising campaigns on Big Data, but there are also applications on financial services (making sure our pension funds take the right risk), engineering, telecom and elsewhere that help improve our lives.
Reducing retail waste is better for everyone. Though some of the stuff they do is questionable, like how carefully items in the store are placed to maximize the amount of unnecessary crap they sell to impulsive people.
Why not grab some public domain data sets and build something on that? There's some really nice sources that can be discovered at data.gov and other places. Much of it could be used to inform and educate the public. You probably could build a small business or consultancy if you choose correctly.
